Author: Alexander Moradi

Appendix Table A12 can be reproduced by running the <Appendix Table A12_Africa.do> file. We created two datasets that replicated Nunn's (2014) test strategy using data from the Afrobarometer round 5.
The <replicating_nunn_ghana.dta> file is different from the <replicating_nunn_africa.dta> file, because the Ghana 0.1 x 0.1 grid did not overlap with the Africa 0.1 x 0.1 grid.

Input files:
1. replicating_nunn_ghana.dta
2. replicating_nunn_africa.dta

Output files:
1) <Appendix Table A12_Africa.xls> (with the results for Africa)
2) <Appendix Table A12_Ghana.xls> (with the results for Ghana)



In order to replicate the datasets, one needs the data from the Afrobarometer Round 5 with GPS coordinates. These data are proprietary. 
We used the <create_replicating_nunn_africa> and <create_replicating_nunn_ghana> do files to create the two data sets. In order for the do file to work, the Afrobaromater data <Merged R5.fulldata.dta> needs to be saved in the same directory.  

The basic steps were the following
1. Using ArcGIS we identified the gridcells within 25 of which the respondent (Afrobarometer) is located
2. Using ArcGIS we identified the gridcells in which the ethnic group (Murdock map) is located
Using the our grid data base we then replicated the variables as closely as in Nunn (2014) 

Input files:
1. <Merged R5.fulldata.dta> (Afrobarometer Round 5 with GPS coordinates)
2a) <grid_africa.dta> (our Africa grid data that contains standard and our controls for Africa)
2b) <grid_ghana.dta> (our Ghana grid data that contains standard and our controls for Africa)
3. <rail_africa_yrbuilt.dta> (which contains data about the date the railroads were built)
4. <ethnic.dta> (data collapsed and aggregated by ethnic group)

Output files:
1. <replicating_nunn_ghana.dta>
2. <replicating_nunn_africa.dta>